ChatGPT Lockdown Mode Rolls Out to All Users

OpenAI is expanding Lockdown Mode, its high-security setting, to all personal and self-serve business ChatGPT accounts. This optional feature was previously only available for enterprise plans. Lockdown Mode lets you dial down ChatGPT’s connected features—like live web access, image generation, Agent Mode, and file downloads—to minimize the risk of prompt injection attacks. In these attacks, a third party tricks the AI into leaking sensitive data or following harmful instructions. When Lockdown Mode is on, browsing uses cached content only, and features that reach outside the conversation are restricted. You can enable it from Settings > Security. OpenAI is also adding an “Elevated Risk” label to certain capabilities in ChatGPT, ChatGPT Atlas, and Codex to help users spot potential vulnerabilities. This move is part of a broader strategy to give users more control as AI gets more powerful and connected. Why it matters: As AI agents tackle real-world tasks involving external services, this trade-off between functionality and security becomes essential—especially for executives or teams handling sensitive information who can’t afford a leak.